Forest change monitoring of a remote biosphere reserve

Abstract

A forest monitoring framework using Universal Transverse Mercator (UTM) grid cells to report forest change estimates derived from time-series satellite imagery was established for the Maya Biosphere Reserve (MBR) in northern Guatemala. Five dates of Landsat Thematic Mapper imagery were acquired and digitally processed to quantify forest change for four time periods: 1986 to 1990, 1990 to 1993, 1993 to 1995, and 1995 to 1997. Time-series change estimates are reported for 215 UTM grid cells approximately 100 km2 each. For the period 1990 to 1997, after the designation of the MBR, the percentage of grid cells with detectable annual forest clearing increased from 38% (1990-93) to 41% (1993-95) and 45% (1995-97). Prior to the establishment of the MBR (1986-90), none of the grid cells exhibited greater than 4.0% annual forest clearing. However in the next three time periods, 7.0%, 8.8% and 9.3% of the grid cells had clearing rates exceeding 4.0% per year. The accuracy of detecting forest clearing was 86.5% over all time periods (Kappa 0.82). Estimates of forest change and user's and producer's accuracy are reported for each time period between 1990 and 1997. The time-series forest change and spatial arrangement of grid locations indicate hot spots where rates and trends of agricultural expansion can be monitored. The baseline survey and the establishment of the UTM grid network to localize forest change estimates provides a framework for future satellite estimates of forest and land cover conversion to be monitored through time. The UTM grid is proposed as the first level in a multi-level ecological monitoring system for the Maya Biosphere Reserve where there are few permanent landmarks in the remote forest region.